Forecasting Trachoma Control and Identifying Transmission-Hotspots

Abstract

Background: Tremendous progress towards elimination of trachoma as a public health problem has been made. However, there are areas where the clinical indicator of disease, trachomatous inflammation-follicular (TF), remains prevalent. We quantify the progress that has been made, and forecast how TF prevalence will evolve with current interventions. We also determine the probability that a district is a transmission-hotspot based on its TF prevalence (ie, reproduction number greater than one).

Methods: Data on trachoma prevalence come from the GET2020 global repository organized by the World Health Organization and the International Trachoma Initiative. Forecasts of TF prevalence and the percent of districts with local control is achieved by regressing the coefficients of a fitted exponential distribution for the year-by-year distribution of TF prevalence. The probability of a district being a transmission-hotspot is extrapolated from the residuals of the regression.

Results: Forecasts suggest that with current interventions, 96.5% of surveyed districts will have TF prevalence among children aged 1-9 years <5% by 2030 (95% CI: 86.6%-100.0%). Districts with TF prevalence < 20% appear unlikely to be transmission-hotspots. However, a district having TF prevalence of over 28% in 2016-2019 corresponds to at least 50% probability of being a transmission-hotspot.

Conclusions: Sustainable control of trachoma appears achievable. However there are transmission-hotspots that are not responding to annual mass drug administration of azithromycin and require enhanced treatment in order to reach local control.

Keywords: control; elimination; forecast; surveillance; trachoma.

Figure 1.

Distribution of district-level prevalence of TF as recorded by the GET2020 Alliance. Each panel represents a different year of data collection. Bin heights represent the number of health districts sampled that had a TF prevalence within the 5% range of each bin. The black line indicates the WHO TF < 5% threshold for district-level control. Colors represent the type of survey with baseline occurring before MDA, impact occurring soon after MDA, and surveillance used to confirm local control.

Figure 2.

Mean TF prevalence of all surveyed districts (top) and percent of surveyed districts with TF less than 5% (bottom). Median estimates and 95% confidence intervals are shown, as determined by bootstrap analysis. Error bars with closed circles represent results from sampling retrospective data with replacement. The error bars with open circles are based on probabilistic forecasts. The horizontal lines indicate an average TF of 5% (top) and a control target of having 90% of districts with TF < 5% (bottom).

Figure 3.

The probability that a district is a transmission-hotspot, as a function of a single TF prevalence measurement. Transmission-hotspots are defined as having a reproduction number > 1 despite annual MDA. Districts with TF values lower than those shown correspond to regions where trachoma transmission is likely to be self-limited. 95% confidence intervals are shown in Supplementary Figure 4.